Three different databases including MEDLINE (PubMed), EMBASE, and the Cochrane Library were systematically searched from their inception to April 2017 by using common keywords related to omega-3 fatty acids and endocrine-related gynecological cancers. The keywords for literature searching were listed as follows: “omega-3 fatty acid,” “fish oil,” “eicosapentaenoic acid,” “alpha-linolenic acid,” “docosahexaenoic acid,” “docosapentaenoic acid” for exposure factors; “endometrial cancer,” “uterine cancer,” and “ovarian cancer” for outcome factors. The bibliographies of relevant studies were also reviewed to identify additional publications. The languages of publication were not limited.

The following are eligibility criteria for individual studies included in the meta-analysis: observational epidemiological studies such as case-control studies and prospective or retrospective cohort studies; studies that investigated the associations between dietary intake of omega-3 fatty acids and the risk of endocrine-related gynecological cancer. For studies using the same data, the more comprehensive study or the first published one was included in the final analysis. Based on the eligibility criteria, two investigators (Tung H and Thu Thi P) independently selected the potential studies.

We assessed the methodological quality of the included studies based on the Newcastle Ottawa Scale for observational studies [23]. The Newcastle Ottawa Scale consists of 3 subscales such as selection of studies, comparability, and exposure. Its star system ranges between 0 and 9. In our study, a study given more than a mean score in each study type was considered as having high quality.

In the main analysis, we investigated the association between dietary intake of overall omega-3 fatty acids (highest versus lowest intake) and the risk of endocrine-related gynecological cancer. Subgroup meta-analyses were performed by type of study design (case-control study or cohort study), type of cancer (endometrial cancer or ovarian cancer), type of omega-3 fatty acids (ALA, EPA, DHA, or DPA), and methodological quality (high or low).

Bidoli et al. [14] and Tavani et al. [15] used the same dataset and transformed into the consumption of ALA and total omega-3 fatty acids intake in risk of ovarian cancer, respectively. To avoid overlaps, study of Bidoli et al. [14], which omega-3 fatty acids were performed much more specific, was only included in subgroup analysis. Besides, Tavani et al. [15] did not show the result for total omega-3 fatty acids and this study was also assessed for subgroup analysis by type of omega-3 fatty acids only.

We used adjusted odds ratios, relative risks, or hazard ratios (HRs) with 95% confidence intervals (CIs) from individual studies in order to calculate a pooled effect size. To measure heterogeneity across studies, we used Higgins I² [24], which is calculated as the following formula:

I²=100%×(Q–df)/Q

, where Q is Cochran’s heterogeneity statistic and df is the degrees of freedom. Negative values of I² are set at zero; I² ranges from 0% (no heterogeneity) to 100% (maximal heterogeneity). If I² value is greater than 50%, it represents the substantial heterogeneity [25]. Because individual studies were conducted in different populations, the random-effects model with the DerSimonian and Laird method was used to calculate the pooled effect size [26,27].

Publication bias was assessed by using the Begg’s funnel plot and Egger’s test [28]. If the funnel plot is asymmetric or the p-value for Egger’s test is lower than 0.05, there exists publication bias. The Stata SE ver. 14.0 software (StataCorp., College Station, TX) was used for all statistical analyses.

Fig. 1 illustrates a flow diagram to identify relevant studies. A total of 3,531 articles were obtained from three databases. Among them, 375 duplicate articles were excluded. After reviewing a title and abstract of each article, we excluded 3,112 articles that did not satisfy selection criteria. Among them, 34 articles were excluded after reviewing the full texts of the remaining 44 articles. The reasons for exclusion were not relevant to study topic (n=9) and insufficient data for study outcome (n=25). A total of 10 studies with six case-control studies [14-17,19,22] and four cohort studies [13,18,20,21] were included in the final analysis.

The general characteristics of the nine studies included in the final analysis are summarized in Table 1. The included studies were five case-control studies with a total of 12,523 participants consisting of 5,279 cases and 7,244 controls, which were published between 2002 and 2014 and four cohort studies with a total of 237,714 participants, which were published between 2002 and 2016. They were conducted in the United States (n=6), Italy (n=2), and Australia (n=1). The follow-up periods ranged between 1980 and 2013.

Table 2 shows the methodological quality of all the included studies based on the Newcastle Ottawa Scale. All the included studies were awarded 7 or 8 stars: three out of five case-controls studies and two out of four cohort studies were given 8. The mean score was 7.6 for case-controls studies and 7.5 for cohort studies.

As shown in Fig. 2, compared to lowest intake of dietary omega-3 fatty acids, highest intake was not associated with the risk of endocrine-related gynecological cancer in the random-effects meta-analysis of case-controls studies (n=5) and cohort studies (n=3) (pooled odds ratio [pOR]/HR, 0.87; 95% CI, 0.73 to 1.04; I²=67.2%). In the meta-analysis by type of study, dietary intake of omega-3 fatty acids was not associated with risk of endocrine-related gynecological cancer in cohort studies (pHR, 1.03; 95% CI, 0.63 to 1.67; I²=81.9%; n=4), while a significantly decreased risk was found in case-control studies (pOR, 0.81; 95% CI, 0.67 to 0.98; I²=55.7%; n=5). No publication bias was observed in the main analysis: the Begg’s funnel plot was symmetrical, and the p for bias from the Egger’s test was 0.41 (Fig. 3).

Further, there was no significant association between dietary intake of omega-3 fatty acids and endocrine-related gynecological cancer in the subgroup meta-analyses by study quality (low vs. high), type of omega-3 fatty acids (ALA, EPA, DHA, and DPA), and type of cancer and type of study in each type of omega-3 fatty acids (Table 3, Fig. 4).

The current study found that there was no significant association between dietary intake of omega-3 fatty acids and the risk of endocrine-related gynecological cancer in the meta-analysis of cohort studies, while dietary intake of omega-3 fatty acids was associated with the decreased risk of endocrine-related gynecological cancer in case-control studies. These findings imply that there is no higher level of evidence to support the preventive effect of dietary intake of omega-3 fatty acids on endocrine-related gynecological cancer such as endometrial cancer and ovarian cancer.

The discrepancies in the effect of dietary intake of omega-3 fatty acids on the risk of these cancers between case-control studies and cohort studies might be associated with some important biases [29]. In general, case-control studies are more sensitive to selection bias and recall bias than prospective cohort studies [29]. Either a case group or control group might not represent the whole population because any group was a non-random sample from the population. This could lead to selection bias. Also, cancer patients might recall their dietary differently from controls: for example, they might answer that they had consumed less foods like fatty fish rich in omega-3 fatty acids. On the contrary, healthy people tend to report the healthy dietary habit, which may overestimate the true effect [30].

Our findings are consistent with those from the previous meta-analyses on the similar topic. Qiu et al. [31] reported that polyunsaturated fat intake was not associated with the risk of ovarian cancer in the meta-analysis of case-control studies and cohort studies. Also, Zhao et al. [32] found that there was no association between polyunsaturated fatty acid intake and the risk of endometrial cancer in the meta-analysis of case-control studies and cohort studies.

There are several hypotheses regarding the potential protective effect of omega-3 fatty acids on endocrine-related gynecological cancer. A nested case-control study reported that increasing concentrations of CRP, which is a marker of chronic systemic inflammation, were associated the increased risk of ovarian cancer [5]. Another nested case-control study also suggested that CRP levels were positively associated with the risk of endometrial cancer [4]. It has been addressed that there are three overarching mechanisms regarding the pleiotropic anti-inflammatory and immunosuppressive properties of omega-3 fatty acids such as EPA and DHA: modulation of nuclear activation (e.g., nuclear factor-κB), suppression of arachidonic acid-cyclooxygenase-derived eicosanoids, and alteration of the plasma membrane micro-organization (lipid rafts) [33]. Regarding direct anti-carcinogenic mechanisms, previous preclinical models demonstrated that omega-3 fatty acids can reduce tumor cell proliferation, migration, and promote tumor cell apoptosis by inhibiting the mechanistic target of rapamycin complex 1 and 2 signaling, which is one of the major targets for the treatment of endometrial cancer [34]. Also, it has been reported that omega-3 fatty acids had anti-proliferative and anti-carcinogenic effects on epithelial ovarian cancer cell lines [35,36]. However, those anti-cancer effects of omega-3 effects were not observed in our meta-analysis of observational epidemiological studies.

To the best of our knowledge, this is the first meta-analysis which reports the associations between dietary intake of omega-3 fatty acids and the risk of endocrine-related gynecological cancers. Our study has several limitations. Firstly, we included a relatively small number of individual studies with six case-control studies and four cohort studies. Thus, further large prospective cohorts studies are warranted to confirm our findings. Secondly, our findings should be limited to dietary intake of omega-3 fatty acids. We planned to evaluate the effects of omega-3 fatty acid supplements on the risk of endocrine-related gynecological cancer. However, when we searched three core databases (PubMed, EMBASE, and Cochrane Library), only VITAL study [18] reported the result which source of omega-3 fatty acids is from diet plus supplement. Thirdly, among studies included in the analyses, only Brasky’s study takes into consideration of dietary factors as adjustment variables. Lastly, our results were not able to be applied to Asians because published data originated from the Western countries, mainly the United States, and different methods of fish cooking between Western and Eastern countries and frequent intakes of raw fish in Far Eastern countries such as Japan and Korea as a source of dietary omega-3 fatty acids may affect the risk of endocrine-related gynecological cancer.

In conclusion, the current meta-analysis of observational studies suggests that there was no higher level of evidence to support the protective effect of dietary intake of omega-3 fatty acids on the risk of endocrine-related gynecological cancer such as endometrial cancer and ovarian cancer. Further larger prospective studies should be necessary to confirm our findings.